Honey, They Shrunk the Processor
Film buffs may vaguely recall “Fantastic Voyage,” a 1966 sci-fi B movie involving a team of intrepid doctors who were magically miniaturized and injected into the bloodstream of a critically ill diplomat to conduct lifesaving microsurgery.
The movie made more of an impression for its depiction of antibodies giving a jumpsuit-clad Raquel Welch a crushing squeeze (having mistaken her for an invading bacterium), than for its technical prescience. But a study by scientists at Yale and Rice universities, recently published in Science magazine, suggests that the movie may have been on to something after all.
The researchers built a simple electrical circuit that can be switched on and off--using the same binary method of managing digital data used in today’s microprocessors, the brains of all electronic devices. But their circuit had a profound difference. It was built on a molecular scale and was thousands of times smaller than the already microscopic circuits found in typical silicon chips.
If molecular-scale processors are perfected, within a decade they could force us to rethink all of our assumptions about the role of computing in everyday life.
Molecular computers would offer the possibility of ubiquitous supercomputing--they would be so small that they could be woven into the threads of your shirt, using your body heat or ambient light for power.
To understand why this could alter the face of computing, consider the daunting challenges now facing makers of traditional silicon-based microprocessors:
For more than three decades the chip industry has advanced in accordance with “Moore’s Law”--the brainchild of Intel co-founder Gordon Moore. Moore said in 1965 that chips’ performance, and hence computing power, would double roughly every 18 months.
Moore’s Law has proved remarkably accurate. Breakthroughs have cropped up with amazing regularity. An example is the announcement last week by UC Berkeley researchers that they had developed a transistor some 400 times smaller than today’s versions.
But even if manufacturers keep overcoming technology barriers and keep circuitry shrinking while performance increases, their efforts could be rendered impractical by Moore’s second law. That precept holds that as chip circuits shrink, the cost of the fabrication plants to build those chips rises dramatically.
Today, at several billion dollars apiece, only a handful of companies can afford to build a new chip “fab.” By some estimates, the cost of a new plant could hit $50 billion a decade from now. If correct, that could make profitable production unfeasible even for Intel, the world’s dominant supplier.
Though each new generation of microprocessor has offered a performance improvement of 10% to 100%, the molecular-scale chips suggest there could be a leap in computing power by a factor of millions or even billions.
Perhaps more important, many researchers believe molecular circuits can be produced at a tiny fraction of the cost of today’s vastly complex technology used to create silicon-based chips. That’s because the circuits built by the Yale-Rice team (as well as those created by Hewlett-Packard with UCLA chemist James Heath) use a purely chemical, or “self-assembly,” process, similar to growing a crystal.
Chemical production could create generic chips that would be many times less complex than today’s processors. Molecular-scale chips would make up for that lack of specialized design with virtually unlimited raw computing power.
To be sure, scientists have taken only the first steps toward that goal. They still must learn how to link millions or billions of such circuits into a workable device and to manufacture them on a mass scale.
Mark Reed, a Yale engineering professor and coauthor of the recent study in Science, said he and his colleagues will give further hints at how molecular switches can be combined to build such a prodigious memory device on Dec. 6 at the International Electron Device Meeting in Washington. Such advances could lead one day to video players that store 10 feature films in the space of a credit card.
If that example sounds strange (who would want to view a movie on a credit card?), it’s because the future devices based on molecular-scale circuitry would basically render our conventional ideas about computing obsolete.
But what would people do with such tiny, powerful computers? “Fantastic Voyage” offers one hint. Phil Keukes, a pioneer in molecular-scale computing at Hewlett-Packard Labs in Palo Alto, believes this technology could create miniature diagnostics that could be shot into the bloodstream and determine, say, what strain of a disease a patient has, then monitor the progress of treatment and feed back a constant stream of data to doctors or shoot micro-doses of drugs directly into diseased cells.
Mark Weiser, the late chief scientist of the Xerox Palo Alto Research Center, imagined a world where “smart paint”--a pigment solution containing millions of micro-machines--could be used to spread stereo speakers on your living room wall. He envisioned a “smart toner” that would reorganize itself limitlessly on a single sheet of paper, creating a book that could be folded into a shirt pocket.
Of course, it’s impossibly hard to foresee what will be done with such tiny and powerful machines.
“If we can truly make this kind of technology manufacturable . . . we’ll have computing that’s cheap enough to throw away,” Reed said. “If you can buy the guts of a computer for 1 cent, how would that change your life?”
Molecular-scale processors “are going to revolutionize something,” he said, “but we don’t know what that something is yet.”
*
Times staff writer Charles Piller can be reached at charles.piller@latimes.com.
